时序图节点嵌入策略的研究

相较于传统的图数据分析方法，图嵌入算法是一种面向图节点的新型图数据分析策略.其旨在通过将图节点向量化表达，进而在节点向量基础上利用神经网络相关技术更有效的进行图数据分析或挖掘工作，如在节点分类、链接预测及交通流预测等经典问题上取得效果显著.虽然研究者们在图嵌入方面已取得了诸多成果，但是面向时序图的节点嵌入问题却未被充分重视，本文便是在先前研究工作的基础上，结合信息在时序图中的传播特性，提出了一种对时序图节点进行自适应嵌入表达的方法ATGEB （Adaptive Temporal Graph Embedding）.首先，为了解决不同类型时序图节点活跃程度不同的问题，通过设计一种自适应方式对其活跃时刻进行聚类.而后，在此基础上设计一种游走模型用以保存节点对之间的时间关系，并将节点游走序列保存在一种双向多叉树上进而可以更快速的得到节点时间相关的游走序列.最后，在基于节点游走特性和图拓扑结构的基础上，对节点向量进行重要节点采样，以便在尽可能短的时间内训练出满足需求的网络模型.通过充分的实验证明，本文面向时序图的嵌入策略相较于现流行的嵌入方法，在时序图时序中节点间时序可达性检测以及节点分类等问题上得出了更好的实验效果.

Node Embedding Research over Temporal Graph

Compared with the traditional graph data analysis method, graph embedding algorithm provides a new graph data analysis strategy. It aims to encoder graph nodes into vectors to perform graph data analysis or mining tasks more effectively by using neural network related technologies. And some classic tasks have been improved significantly by graph embedding methods, such as node classification, link prediction, and traffic flow prediction. Although plenty of works have been proposed by former researchers in graph embedding, the nodes embedding problem over temporal graph has been seldom studied. This paper proposed an adaptive temporal graph embedding, ATGED, attempting to encoder temporal graph nodes into vectors by combining previous research works and the information propagation characteristics together. First, an adaptive cluster method was proposed by solving the situation that nodes active frequency is different in different types of graph, and then design a new node walk strategy in order to store the time sequence between nodes, an also the walking list will be stored in bidirectional multi-tree in walking process to fast get complete walking lists. Last, based on the basic of walking characteristic and graph topology, we proposed an important node sampling strategy to train the satisfying neural network as soon as possible. Sufficient experiments demonstrating out methods surpass existing embedding methods on node clustering, reachability prediction and node classification in temporal graphs have been conducted in this paper.